Show HN: Aflow achieves 96.2% on HumanEval at 4.55% of GPT-4's cost

We've developed AFLOW, an automated agentic workflow generator using Monte Carlo Tree Search:

Outperforms human-designed workflows in Coding, Math, and RAG Achieves GPT-4o-level performance on HumanEval at just 4.55% of its cost 96.2% HumanEval accuracy using GPT-4o with AFLOW Generates custom workflows in 1.5h with just an eval function